Dec 15, 2007

FLL Mission Overview - Summary

Well, this FLL season is just about at its climax: some teams have already had their tournaments, while others will have them in the following weeks. This was an interesting year, with several new features coming into play in the competition. We hope our Mission Overview posts have been helpful to teams... here are our comments on the season in general, and follwing them is a list of links to all the posts for convenience.

Jonathan's Comments:

I think the major difference in this year's challenge was that there was a larger number of missions that were relatively easier to solve. So while each individual mission was, on average, easier to solve than last year's, it was harder to get many of them done in the time limit. Apparently the challenge was made such that a perfect score would be very rare (the designer, Scott Evans, said he considered that if a team got a perfect score, they "beat" him, because he tried to make it really hard to get one), but it may well have turned out to be easier than last year.

One of the new aspects of the teams this year is the amount of NXT robots vs. RCX robots being used. Out of about 60 teams I saw in NJ, I only noticed a few that used RCX robots. This is interesting since the fairness bonus for using an RCX is quite a bit better than last year's. The NXT is definitely catching on in the FLL community.

Richard's Comments:

Though I haven't been to an official tournament, a scrimmage practice with 10 teams I recently attended had all NXT robots! And since most of these teams were new, I think the NXT is really causing people to start up teams.

I think this is one reason Scott Evans might have made it easier than claimed; he guessed that the release of the NXT would start many new teams, so making slightly simpler missions would make these new teams' first year smoother, while the complexity of other missions would still give (somewhat) a challenge to older teams.

And even though teams are scoring higher, I've noticed that robots are no longer as technical (? correct word?); many rely on dead-reckoning (even though teams like Jonathan's shows that this works just as well) and no longer use many or any sensors at all. In the days of the RCX, teams had to rely on sensors to get any accuracy at all. I think this is one area that the NXT can improve in.

So good luck to everyone, and I hope to see some of you at the World Festival in Atlanta!

Links:


-Jonathan and Richard

19 comments:

CATpit contractor said...

I share Richards comments regarding the "dumming" down of robot techniques. Younger teams might not know how to do any of that stuff, more experienced teams might be struggling with a new programming language for the NXT, and still other teams (like Jonathan's) have simply figured out that if you do things right, it doesn't matter how you program the robot.

My team took six years to figure out all of the programing tricks we used on this year's robot (and we use a lot!). Newer teams just don't have that kind of experience, and, as Richard pointed out, the NXT is gathering lots of new teams (which is a good thing).

Of course, then you have team's like Jonathan's... I can't relate:)

Jonathan Daudelin said...

Lol... well the missions in FLL were all made such that they could be solved without any sensors besides rotation sensors, so the team figured they might as well do it simpler rather than more complicated (not to say nobody should use other sensors, of course - different teams find different solutions that work best for them).

-Jonathan

tr6scott said...

I agree with Richard's comments too. In our November warm up match we could do all the missions in 5 minuets. The team had to scrap some missions and resequenced other to get the most points the night before. The team worked hard on finding programming and hardware solutions to make the changeovers quicker, along with rehearsing in 2 man teams at the board, as opposed to tagging in the whole team at each match. When everything was said and done, at regionals we had beat Scott's game in time, but the perfect score alluded us by the solar panel. 385 then at States no perfect scores by us on the completion tables, but many were run in the pits.. We ended up 4th overall. There was a 400 posted with an NXT and a 399 posted with an RCX. (Novi Michigan State)

Matthias Paul Scholz said...

Jonathan,

First of all, thanks for tremendeous work you have done to sort all these missions out - very helpful!

As to your statement
"In the days of the RCX, teams had to rely on sensors to get any accuracy at all. I think this is one area that the NXT can improve in."

I guess that hasn't been a matter of the NXT but of the way the missions have been shaped this year (you touched on that already): according to my experiences on FLL 2007 events I visited, it was more about precise placing and targeting the robot by the team member at the base than navigation by the robot. In a lot of cases, you actually could have used a windup machine instead of a programmable robot.

Joshua Heinzl said...

"still other teams (like Jonathan's) have simply figured out that if you do things right, it doesn't matter how you program the robot."

I disagree and think that programming is one of the most important parts. Lego parts are limiting and not perfect because it is a toy. However, good programming can reduce or eliminate many limitations.

Dumbing down is only happening because teams think that nxt-g's built in move block is good enough and the tribot works well.

"In the days of the RCX, teams had to rely on sensors to get any accuracy at all. I think this is one area that the NXT can improve in."

I think this is an area TEAMS can improve in - it has nothing to do with RCX/NXT. As I recently commented on another post, the limitation is TEAMS thinking odometry is good enough. Other techniques are necessary to reliability and repeatability.

Josh

Anonymous said...

Of course you disagree with Jonathan.

What a surprise.

;P

Jonathan Daudelin said...

Actually, I agree that programming is an important part of the robot... not more important than building, but not less important either - your robot won't do anything unless it has both. :P I've just found that good programming doesn't necessarily mean complicated programming. :-)

-Jonathan

Joshua Heinzl said...

I was not disagreeing with Jonathan, rather with catpit. If everyone agreed about everything no progress would every be made because improvements would not exist. It is people's different views that the world revolves around.

With some robots programming is more important than with others.

I personally have limited experience with NXT-G. Would Jonathan or anyone else like to comment on how what makes 'good' programming with NXT-G? Does everyone just use the move blocks or are you using more complicated routines?

Josh

Jonathan Daudelin said...

Josh,

I think of good programming as programming that makes the robot complete its task successfully. The more efficient, fast, and simple the program is, the better.

-Jonathan

Joshua Heinzl said...

I agree. We would add the twist that reliability under ALL circumstances is a requirement, which you may have also implied.

Josh

Jonathan Daudelin said...

Sort of... reliability in *ALL* circumstances doesn't necessarily make the program better (perhaps it doesn't need to work in all circumstances), but consistency is definitely a factor.

-Jonathan

Joshua Heinzl said...

Well a reasonable list of circumstances would have to be defined. Most FLL robots won't work on a rocky surface but they should be able to work on various mat conditions, battery voltages, etc.

Josh

Anonymous said...
This comment has been removed by a blog administrator.
Anonymous said...

Jonathan noticed that most teams were using NXTs. No one mentioned the fact that new teams this year were forced to use NXTs, they could not even purchase older RCX kits.

This, combined with the fact that older, more experienced teams are probably more likely to move to the NXT for various reasons and you'll see why there weren't many RCX teams.

I do think the bonus this year was far too generous. 60 points is a lot, and mapping scores like 350 to 390 and 375 to 395 really puts competitive teams using NXTs at a big disadvantage.

It also has the unfortunate result of putting first year teams at a disadvantage since they're limited to NXTs.

I think going forward there should be no bonus. It has been two years now and anyone wanting to change can change.

Jonathan Daudelin said...

anonymous,

Actually, it's very easy to buy the RIS on EBay, and I'd expect you could get it other places as well.

-Jonathan

Joshua Heinzl said...

But most rookie teams would have ordered the NXT Robot Kit when signing up. Yes, if once the rules were released they decided that an RCX was necessary they could have gone to eBay, but this isn't so likely.

Josh

Anonymous said...

Hey everybody look!!!! Josh disagrees with Jonathan. What a surprise!!!!! ;P

Joshua Heinzl said...

It is a brave man who hides behind the anonymous cover.

If everyone agreed on everything, then improvements would never be made. The world revolves around people's varying viewpoints.

If you would like to contribute to this discussion constructively with your real name I would be happy to continue.

Josh

Jim Kelly said...

Both of you, Anonymous and Josh, cut it out.

Jim

Related Posts Plugin for WordPress, Blogger...